Portal de Periódicos da CAPES

Is mortality among under-five children in Nairobi slums seasonal?

2009; Wiley; Linguagem: Inglês

10.1111/j.1365-3156.2009.02419.x

1365-3156

Maurice Mutisya, Benedict Orindi, Jacques Emina, Eliya M. Zulu, Yazoumé Yé,

Poverty, Education, and Child Welfare

Objective To investigate the seasonal pattern of overall mortality among children aged below 5 years living in two informal settlements in Nairobi City. Methods We used data collected from January 2003 to December 2005 in the Nairobi Urban Health and Demographic Surveillance System on demographic events (birth, death, and migration). Analyses of seasonal effects on under-five mortality are based on Poisson regression controlling for sex, age, study site and calendar year. Results During the study period, there were 17 878 children below 5 years in the study sites. Overall 436 under-five deaths were recorded. The overall death rate for the under-five children was 19.95 per 1 000 person years. There is a significant seasonal variation of under-five mortality. The mortality risk was significantly higher in the second and third quarters of year than in the fourth quarter (RR = 1.6, CI: 1.3–2.2 and RR = 1.5, CI: 1.1–2.0). Conclusion This paper demonstrates that overall mortality among under-five children in the urban poor is seasonal. Overall during the second quarter of the year, the death rate increases by nearly twofold. This evidence generated here may help to support well targeted interventions in reducing under-five mortality in the slums. Objectif: Investiguer les profils saisonniers de la mortalité globale chez les enfants âgés de moins de cinq ans vivant dans deux quartiers informels à Nairobi City. Méthode: Nous avons utilisé des données collectées entre janvier 2003 et décembre 2005 dans le Système de Surveillance Démographique et de Santé Urbaine de Nairobi sur les événements démographiques (naissances, décès et migration). Les analyses des effets saisonniers sur la mortalité des moins de cinq ans sont basées sur la régression de Poisson avec ajustement pour le sexe, l'âge, le lieu de l'étude et l'année. Résultats: Durant la période d'étude, il y avait 17878 enfants de moins de cinq ans dans les sites d'étude. Au total 436 décès de moins de cinq ont été enregistrés. Le taux global de mortalité des enfants de moins de cinq était de 19,95 pour 1000 personnes-années. Il existe une variation saisonnière significative de la mortalité des moins de cinq ans. Le risque de mortalitéétait significativement plus élevé dans les deuxième et troisième trimestres de l'année que dans le quatrième trimestre (RR = 1,6; IC: 1,3- 2,2 et RR = 1,5; IC: 1,1-2,0). Conclusion: Cette étude démontre que la mortalité globale chez les enfants de moins de cinq ans dans les populations urbaines pauvres est saisonnière. En général, au cours du deuxième trimestre de l'année, le taux de décès augmente de près de deux fois. Cette preuve pourrait contribuer à soutenir les interventions bien ciblées pour la réduction de la mortalité des moins de cinq ans dans les bidonvilles. Objetivo: Investigar los patrones estacionales de la mortalidad entre niños menores de cinco años viviendo en dos barriadas de la ciudad de Nairobi. Métodos: Hemos utilizado datos recolectados entre Enero 2003 y Diciembre 2005 a través del Sistema de Seguimiento Demográfico y de Salud Urbana de Nairobi, sobre eventos demográficos (nacimientos, muertes y migraciones). Los análisis de los efectos estacionales sobre la mortalidad en menores de cinco años están basados en la regresión de Poisson, controlando para sexo, edad, lugar de estudio y año. Resultados: Durante el periodo del estudio había 17,878 menores de cinco años en los lugares del estudio. En total se reportaron 436 muertes de menores de cinco años. La tasa de mortalidad para este grupo de edad fue de 19.95 por 1000 personas año. Hay una variación estacional significativa en la mortalidad de menores de cinco años. El riesgo de muerte era significativamente mayor en el segundo y tercer trimestre del año, que en el cuarto (RR=1.6, CI: 1.3; 2.2 and RR=1.5, CI: 1.1; 2.0). Conclusión: Este artículo demuestra que la mortalidad total de los menores de cinco años entre urbanitas pobres es estacional. En general, durante el segundo trimestre del año, la tasa de mortalidad aumenta casi al doble. Esta evidencia podría ayudar a la hora de diseñar intervenciones para reducir la mortalidad de los menores de cinco años viviendo en barriadas. Childhood mortality remains high in developing countries, especially in sub-Saharan Africa (SSA). Despite the observed decline in the past decade (Black et al. 2003) progress is slow in many SSA countries (WHO 2000). Close to five million under-five deaths continue to be recorded in SSA each year and account for close to half of the world's under-five deaths. Actually, the emerging evidence that child mortality is either increasing or not declining as much as expected in many African countries (Garenne & Gakusi 2006) raises serious concerns about the continent's capacity to achieve the MDGs related to child health. Moreover, growing urban populations in developing nations are posing new challenges in efforts to improve child survival. The result of rapid urbanization amidst poor economic performance and governance of urban areas is sprawling slums in cities where the majority of urban populations currently reside (United Nations Human Settlements Programme (UN-Habitat). 2003, Ooi & Phua 2007). The settlements are characterized by a poor living environment- poor sanitation and drainage, overcrowding, and lack of social services like health facilities, schools and water supply (African Population and Health Research Center 2002; United Nations Human Settlements Programme (UN-Habitat). 2003; Ooi & Phua 2007). This kind of environment poses health risks to the population but more so to children. Intra-urban differences particularly in child malnutrition and mortality are becoming more evident than rural–urban differences (Fotso 2006). For instance, childhood mortality among the slum residents in Kenya was estimated to be 151 deaths per thousand in 2000, compared to 95 for the whole of Nairobi and 117 among the rural population of Kenya (African Population and Health Research Center 2002). Similar patterns have been reported in other African settings (Madise & Diamond 1995; Madise et al. 2003). The relatively poor health outcomes among children living in poor slum settlements (African Population and Health Research Center 2002; Kyobutungi et al. 2008; Ndugwa & Zulu 2008); the fact that the majority of urban residents live in slum settlements (UN-Habitat 2003); and the projections that more than half of Africans will live in urban areas by 2030 (United Nations 2008) pose a new challenge in global and national efforts to reduce child mortality in Africa. Indeed, with less than 6 years to go to the 2015 landmark for the Millennium Development goals, most African countries are likely to fail to meet their goals related to child mortality if they do not address the worsening health conditions of the urban poor. There is, therefore, a need to better understand the factors associated with this high mortality. However, studying childhood mortality in the urban poor is often challenged by lack of appropriate data. The DHS and the DSS offer a great opportunity to study the determinants of under-five mortality. Using such data, various scholars have documented the effect on childhood mortality of several factors including individual, household and community and environmental characteristics (Ayaga & Zuberi 2003; Amouzou & Hill 2004; Ndugwa & Zulu 2008). While many studies have examined the determinants of mortality including season (Muhuri 1996; Jaffar et al. 1997; Findley et al. 2005) few have studied seasonal patterns of childhood mortality among the uniquely vulnerable urban poor (Ye et al. 2009b). The importance of seasonality on child survival is clearly shown in the conceptual framework by Mosley and Chen (1984) which suggests that child survival chances are due to operation of biological, social economic and environmental forces. In the framework, seasonality is grouped together with climate, rainfall and temperature under the ecological setting of community level factors (Mosley & Chen 1984). These ecological factors affect not only the availability of food, but are said to influence mothers' attention to their children, especially during rainy seasons when quality of water and sanitation conditions become compromised. The effect of season is also seen in the prevalence of diseases such as diarrhea, gastrointestinal and respiratory diseases. For instance, the seasonal pattern of death due to diarrhea, gastrointestinal and respiratory diseases has been observed in the 20th century; whereby peak of respiratory diseases has been observed during the winter season (Apostolidou et al. 1994;Madrigal 1994; Mackenbach 1997). Recently, increases in malaria incidence and mortality have been shown to occur during or after the rainy season in West Africa (Hammer et al. 2006). Furthermore, diarrhoeal diseases are known to peak during rainy and cool seasons. Most of these studies were mostly among the rural population and thus little is known about seasonal effects on childhood mortality among the urban poor. A recent study from the slums of Nairobi shows that there is seasonality in Pneumonia mortality among the under-five children (Ye et al. 2009b). However, it is not evident whether and how this seasonal pattern applies to overall mortality. The poor environmental conditions that affect child survival in the slums change by season, and they are worse during the rainy period. Main childhood diseases like diarrhoea are more prevalent during the rainy season due to greater contamination of the environment and water supply during this period. Studying seasonality of childhood mortality among the urban poor will thus help policy makers and program managers to implement appropriate interventions on reducing child mortality among the urban poor and monitor their effectiveness during high-risk seasons (Kandala et al. 2006). Using the longitudinal Nairobi Urban Health and Demographic Surveillance System (NUHDSS) data, this paper explores whether and the extent to which childhood mortality among the urban poor is seasonal. The NUHDSS, which covers about 60 000 people, is run by the African Population and Health Research center (APHRC) since 2002. The NUHDSS covers Korogocho and Viwandani slum settlements that settle on a land area of about 0.97 km2. Poor environmental sanitation; overcrowded houses and poor access to basic health care characterize these settlements (APHRC 2002, Amuyunzu-Nyamongo & Taffa 2004; Ndugwa & Zulu 2008). Nairobi is situated at a high altitude of 1700 m above sea level. This geographical situation makes Nairobi relatively cold as compared to other tropical cities. The average annual minimum temperature is 12 °C (range: 11–14 °C) and average maximum temperature is 23 °C (range: 21–26 °C). The total annual precipitation is about 900 mm. There are two rainy and two dry seasons. The warm and dry season goes from mid-December to mid-March. The second season is from mid-March to May and corresponds to the main rainy season. The third season goes from June to mid-October and it is cold and dry. The last season correspond to the short rainy season (mid-October to mid-December) (Figure 1). The poor drainage and waste management system in the slums results in contamination of drinking water during the rainy season. Average monthly temperature and rainfall for Nairobi from 2003 to 2005. (Kenya Meterological Department) Data used in this study was for the years 2003 to 2005 and was extracted from the NUHDSS database. All children aged less than 5 years between 1 January 2003 and 31 December 2005 were included in the study sample. Children aged 5 years or older were only observed at the time they were at risk. The sample included 17 878 children born in the study area and those who in-migrated into the study area as long as they were within the age limits. Key demographic events (birth, death, migratory movements) are routinely recorded for all residents in households under the NUHDSS. These data are collected every four months by highly trained and experienced interviewers. Through this system the deaths that occurred in our study population were recorded. Since households are visited three times in a year, this ensures the date of the event is reported correctly. The data quality control procedures include refresher trainings that are usually conducted before the start of a new data collection cycle and regular spot-checks that are done by field supervisors. To determine the actual observed population at different times, we calculated person months of exposure for each month. Age in months was calculated for each month and in years for the different years. A child who survived throughout the 3 years contributed 36 person months (equivalent to 3 years). Deaths occurring during the study period were identified by month and year of death to determine whether there was any seasonal pattern. The number of deaths occurring per month was calculated and denoted as Dij (where 'i' is month in question and 'j' the year). To allow for meaningful comparison, months were grouped to form four quarters of the year (First quarter: January to March; Second quarter: April to June; Third quarter: July to September and the October to December as the fourth quarter). For calculation of the overall monthly and quarterly mortality rates, deaths occurring in a specific month were summed up for the 3 years (Dij) and divided by the number of person time contributed in that month for the 3 years. Similarly, for year quarterly death rate, deaths occurring during the months forming that quarter for the different years were summed up and divided by total person time contributed during the same quarter. Death rates were calculated as the total number of deaths recorded during the observation period divided by the total person time. Mortality rates were expressed per 1000 person years (PYs). The mortality rates among children aged less than 5 years calculated in this study is different from the typical under-five mortality rate used in demography. The period of observation is 3 years rather than the normal 5 years, which means that a child can only contribute a maximum of 36 months of exposure regardless of the date of birth. The other difference with the typical demography definition of the under-five mortality rate is that we allow children who migrated into the surveillance area to be part of the population at risk for the time that they lived there. The monthly proportion of deaths was calculated by dividing the number of deaths in that month (MAmonth (i)) by total number of children at risk during that month. Finally plotting of the graphs was done using the logarithm of the proportion of death in that month. To estimate absolute and relative risks of dying at different times of the year, data was stratified in terms of sex, year quarter, age group, site, and year of study (in order to control for the effect of entrance into the study population). A Poisson regression was fitted using dummy variables for each of the above variables, with the outcome being the number of deaths in each group. The variables included in the model are presented in Table 1. For the study period (2003–2005), NUHDSS data comprised 17 878 children accumulating 21 805 person years. Contribution of person time between the two sexes was relatively equal, i.e. males contributing 49.4% of the total person time. Similarly, distribution of person time in the 3 years was more or less the same, indicating stability in the number of children aged less than 5 years each year: 32.7%, 32.8% and 34.5% for years 2003, 2004, and 2005 respectively. Children aged 1–4 years contributed 75.5% of the person time (Table 2). During the study period, we observed 436 deaths and the overall death rate was 19.95 deaths per 1000 Person years while infant mortality was 55.17 deaths per 1000 person years. The deaths were distributed by year as follows: 2003 = 156 cases (35.8%), 2004 = 139 cases (31.9%) and 2005 = 141 cases (32.3%). The death rate was 21.9/1000, 19.4/1000 and 18.8/1000 for the years 2003, 2004, 2005 respectively. The quarterly deaths rates show that the mortality rate is relatively high in the second quarter (24.8 deaths per 1000 PYs) of the year (Table 3). This rate is much higher than the rate in the fourth quarter of the year (14.1 deaths per 1000 PYs). The rates for the first quarter (19.9 deaths per 1000 PYs) and the third quarter (21.9 deaths per 1000 PYs) are also quite high, although lower than the second quarter. To check the monthly variation of death rate, the logarithms of the proportion of all deaths were plotted by calendar month. Figure 2 shows logarithm of the proportion of under-five deaths per month plotted against months of the year. The results show that the U5 death rate between April and August is high with a pronounced peak occurring between June and July. By and large, under-five mortality starts to rise from April with a sharp increase between May and July, which coincide with the period of long rains and onset of the cold season in Nairobi. Seasonal pattern of all causes mortality among Nairobi slum children. To examine the adjusted seasonal effect on the risk of childhood death during the period of study a Poisson regression model was fitted controlling for sex, year, site, and age group. Two models were fitted: one for overall under-five mortality and another for infant mortality since the seasonal effect on the risk of dying among the infants is different to that of post infant ages (Table 4). The overall risk of under-five death is significantly higher in the second and third quarters of the year RR: 1.6 (CI: 1.3–2.2) and 1.5 (CI: 1.1–2.0) than in the fourth quarter, which saw the fewest deaths. That is, the risk of death during the second and third quarters was significantly higher than the fourth quarter by 60% and 50%, respectively. The first quarter's mortality was also higher, although the effect was not significant at 5% level. The seasonal patterns in mortality are similar for infants and all under-five, although, the risk of death during the second and third quarters of the year is comparatively high among the under-five than among the infants. The results also indicate that boys are at significantly higher risk of dying than girls for both models. The risk of death among girls falls significantly by 21% overall, and by 23% among the infants. Mortality risks across the 3 years were not statistically different in both models, although risk of dying between 2003 and 2005 reduces by about 20%. Children in Korogocho are at higher risk of dying before the fifth as well as the first birthday than children in Viwandani. The risk of dying was higher in Korogocho than Viwandani by about 64% overall and by 51% among the infants. As expected, the risk of death is significantly lower among children aged 1–4 years (by 84%) than among infants. The risk of childhood mortality in the slums of Nairobi City is high between May and July, increasing significantly by about 50% compared to the rate of death between October and December. This is the period of the long rains and onset of the cold season. The fourth quarter of the year, which experiences onset of short rains, had the least number of deaths. Slums are most unhygienic during the rainy seasons. Deterioration of waste disposal and of sanitation systems, as well as probable contamination of drinking water during the rainy season could explain why mortality rates for children elevate during the second quarter (Muhuri 1996). Waste disposal becomes a more serious hazard during the rainy season because solid as well as liquid waste filters into broken plastic water pipes that water vendors use to make illegal connections to the city council water system (African Population and Health Research Center 2002). Because of the poor drainage and waste disposal systems, it becomes uncomfortable and filthy to walk in the slums after it has rained. Consequently, children spend most of their time during the day in their one-room houses, thereby increasing overcrowding and exposure to carbon dioxide from cooking (Awasthi et al. 1996; African Population and Health Research Center 2002; Emuller et al. 2003). As they play outside, children often come in direct contact with the contaminated waste and flowing water after rains, thereby increasing the chances of contracting water-borne infections. Furthermore, the wetness, poor drainage systems, and disposal of garbage in drainage trenches create a favourable environment for the proliferation of infectious diseases, thereby making slum residents more susceptible to diseases such as diarrhoea, pneumonia, and other respiratory tract infections (Kyobutungi et al. 2008). Recent studies in Gambia and Burkina Faso have shown that infant and childhood mortality significantly rises with increased rainfall, as observed in this study (Hammer et al. 2006). While the increased mortality during the rainy season in the Gambia and Burkina Faso is mostly attributed to more cases of malaria during this period, the slums of Nairobi are different since there is no transmission of malaria in the city due to its high altitude (Ye et al. 2009a). The leading causes of death in the slums of Nairobi that account for more than half of deaths among the under-five children are pneumonia (22.8%), diarrhoea (19.5%) and stillbirths (16.3). Although malaria is among the top 10 causes of death in Kenya, it only accounts for 3.5% of deaths in the slums of Nairobi (Kyobutungi et al. 2008) and as noted above, the malaria cases in the slums are probably imported from rural areas by migrants since there is no transmission of malaria in the city (Ye et al. 2009a). It has been noted in the slums of India that diarrhoea increases the number of deaths during the rainy season and that under-five mortality increases during this period (Gupta & Baghel 1999). This study also shows that under-five mortality is significantly higher in the third quarter (July to September) than in the fourth quarter. This mortality pattern corresponds to the temperature pattern during the year. Mortality is higher during the dry cold seasons and lower during the warm and dry seasons. However, the highest monthly mortality was observed in the month of June which marks the beginning of cold season in our study setting. During this month there is a sudden drop of temperatures with the minimum level ranging from 11 °C to 14 °C (the mean annual temperatures for Nairobi is 24 °C). The high prevalence of acute respiratory infections during the dry cold season could explain the high mortality observed in our study setting (Dowell et al. 2003; Emuller et al. 2003; Tsai et al. 2003; Mohamed et al. 2004). These results are in line with the finding that deaths due to pneumonia are highest during the relatively cold season (third quarter) (Ye et al. 2009b). The findings of this study indicate that policy-makers need to understand not only the drivers of mortality, but also when those drivers are most prevalent in order to do proper targeting of interventions to improve child survival. A recent study showed that while high socioeconomic status does not provide a significant advantage in morbidity, it does have a big impact on seeking formal health care since those with more resources are able to pay for health care when their children get sick from the poor environmental conditions (Ndugwa & Zulu 2008). These findings suggest that interventions to improve child survival should involve strategies to increase access to health care for the poorest people during the times when peak mortality is observed. There is therefore a need to study the seasonality of diarrhoea and other causes of death among the urban poor in our setting. The verbal autopsy method used in the NUHDSS is a great opportunity to explore the seasonality of specific causes of death. Some limitations can be highlighted from our study. The yearly quarters represent changes in seasons; however, we did not link actual changes in temperature and rainfall to under-five mortality. Furthermore, one quarter is three months long and there may be wide variation in the daily temperature over that period which was masked in our analysis and may explain the moderate associations with seasons. The study also assessed the seasonal pattern based on 3-year data. It would be important to examine the pattern over many years to determine whether this is a permanent feature of mortality in the slum settings, since there may be annual variation in weather patterns. Our study confirms that under-five mortality has a seasonal pattern among slum residents in Nairobi City. Children are most vulnerable to dying during the rainy and cold seasons. The findings indicate that proper targeting of interventions to improve child survival among the rapidly growing population of the urban poor should pay particular attention to addressing the main drivers of mortality during the seasons when children are most vulnerable to infections. NUHDSS work is supported by grants from the Rockefeller Foundations and the William and Flora Hewlett Foundation. We wish to acknowledge the APHRC's field, data entry and research team.